Red Grape Waste Contains Natural Compounds That May Block Inflammation and Metabolic Syndrome Enzymes Using an Eco-Friendly Extraction Method

Folin–Ciocâlteu reagent, choline chloride ≥ 98% ((2‐hydroxyethyl)trimethylammonium chloride), lactic acid 90%, sodium phosphate dibasic heptahydrate ≥ 98%, sodium phosphate monobasic monohydrate ≥ 98%, sodium chloride ≥ 99%, 4‐Nitrophenyl α‐D‐glucopyranoside ≥ 99%, linoleic acid ≥ 99%, quercetin ≥ 98% (HPLC), α‐amylase from porcine pancreas (~50 U/mg), α‐glucosidase from Saccharomyces cerevisiae (≥ 10 units/mg protein), lipoxygenase from Glycine max (soybean) (≥ 50,000 units/mg solid) were purchased from Sigma‐Aldrich (Millipore Sigma, Steinheim, Germany). Acarbose, 95% was purchased from Acros Organics (China), 3,5‐Dinitrosalicylic acid (DNS), starch soluble (from potatoes) were purchased from Lach‐Ner. S.r.o (Czech Republic), methanol for HPLC, ≥ 99.9%, was purchased from Honeywell (Seelze, Germany) and sodium carbonate was purchased from S.C. Chemical Company SA (Iasi, Romania).

The unfermented red grape pomace (RGP) ( Vitis vinifera L.) from the Fetească Neagră and Merlot varieties was kindly provided by Bratu winery, located in Odaia Manolache village, Vânători commune, Galați County, Romania (45°33′27.5182″ N, 28°0′21.7552″ E). The raw material was stored at –20°C in refrigerated containers until use. The RGP were preserved by drying in a hot air convection (Stericell 111 ECO Hot AIR Sterilizer; MMM Medcenter Einrichitungen GmbH, München, Germany) at a moderate temperature (35°C) up to a moisture content of 8%. Subsequently, the dried material was ground using an electric grinder (Coffee grinder Heinner HCG‐150SS; Heinner, Network One Distribution, Bucharest, Romania) and stored in vacuum‐sealed bags under refrigeration conditions (4°C).

The UAE using NaDES as solvents was performed at optimum parameters, as explained in our previous study (Balan et al. 2025). In this study, the solvent was represented by a combination of choline chloride, lactic acid, and water in a molar ratio of 1:2:1, whereas the UAE was performed at a temperature ranging from 40°C to 60°C; extraction time between 30 and 60 min; and the solid–liquid (S/L) ratio of NaDES (1:10–1:20 mL). From the data reported earlier (Balan et al. 2025), the optimum parameters for RGP bioactive extraction were: temperature of 60°C, for 60 min of extraction and S/L ratio of 1:10 mL. The models were validated, whereas the extraction was repeated three times in order to exhaust the RGP in bioactive.

The modified Folin–Ciocalteu colorimetric technique, described by Serea et al. (2023), was used to quantify the TPC in the NaDES extract. In brief, 200 μL of diluted extract (1:5) was mixed with 15.8 mL ultrapure water and 1 mL of Folin–Ciocalteu reagent and allowed to react for 10 min. Further, 3 mL of 20% Na₂CO₃ was added, and the mixture was kept in a dark place for 60 min at 25 C. The absorbance at the wavelength of 765 nm was measured using a spectrophotometer (Biochrom; Libra 22 UV/Visible Spectrophotom‐eter; 177 Holliston, MA, USA). Measurements were made in triplicate for each sample, and the results obtained were expressed in milligrams of gallic acid equivalents per mL (mg GAE/mL).

For the determination of phenolic compounds profile by HRMS analysis, an aliquot of the sample was dissolved with 1 mL mixture of water: methanol = 80:20, filtered to 0.45 μm hydrophilic membrane filter and then subjected to the instrumental analysis. The identification and quantification of the phenolic compounds in the extract were performed by using an ultra high‐performance liquid chromatograph UltiMate 3000 (UHPLC) (ThermoFisher Scientific, Bremen, Germany), coupled to a Q Exactive Focus mass spectrometer, Focus Hybrid Quadrupole—OrbiTrap (Thermo Fisher Scientific) equipped with HESI. Chromatographic separation of phenolic compounds was performed on a Kinetex C18 column (100 × 2.1 mm; 1.7 μm particle diameter) maintained at 30°C, under a gradient elution of two mobile phases, A (0.1% formic acid in water) and B (0.1% formic acid in methanol), at flow rates of 0.3 and 0.4 mL/min, as previously optimized (Onache et al. 2022). The mass spectrometer was operated in negative mode, in a range between 75 and 1125 m/z, at a resolution of 70,000, under the previously optimized HESI‐source parameters (spray voltage 2.8 kV, capillary temperature 320°C, auxiliary gas heater temperature 413°C, sheath and auxiliary gas flow (N₂) 48 and 11 arbitrary units) (Onache et al. 2022). Identification and quantification of the phenolic compounds from the extract were performed according to their spectral characteristics: mass spectra, accurate mass, and characteristic retention time against external standard solutions. Fragmentation studies by a data‐dependent scan with collision‐induced dissociation (CID) were performed in order to confirm the identified compounds. Compound quantification was performed based on the external standard method, with calibration standards between 25 and 2000 μg/L for each of the phenolic compounds, obtained by serial dilution with methanol of the 10 mg/L calibration standard mixture. All stock and working solutions were stored in the dark at 4°C. The results were expressed in mg/kg of extract. Xcalibur software (Version 4.1) was used for instrument control, data acquisition and data analysis.

The theoretical and computational predictions for the absorption, distribution, metabolism, and excretion (ADME) as well as the oral and nutritional toxicity of the phytochemical compounds identified in the NaDES extract were performed using the SwissADME online server (swissadme.ch; accessed on 19 February 2026) (Daina and Zoete 2016), and the ProTox‐3.0 online server (tox.charite.de↗; accessed on 19 February 2026) (Banerjee et al. 2024). The encoding molecular structures (SMILES) of the phytochemical compounds were obtained from the PubChem online database (pubchem.ncbi.nlm.nih.gov↗; accessed on 19 February 2026).

The efficiency of the optimized extracts in inhibiting α‐amylase was tested using the protocol described by Mahdi et al. (2020) and Serea et al. (2022) with some modification. A volume of 200 μL of enzyme solution (prepared by dissolving 1 g of α‐amylase from porcine pancreas in 13 mL of 20 mM PBS, pH = 6.9, containing 6 mM sodium chloride) was mixed with 200 μL of different diluted extract solution (concentration expressed in mg GAE/mL) and incubated for 10 min at 37°C. Subsequently, 200 μL of starch solution (0.5% w/v) was added, and the mixture was further incubated for 15 min at 37°C. Further, 400 μL of DNS reagent was added, and the mixture was placed in a boiling water bath for 5 min, followed by rapid cooling under a water jet. The samples were then diluted with 1 mL of distilled water, and the absorbance was measured at 540 nm using a UV–VIS spectrophotometer (Biochrom; Libra 22 UV/Visible Spectrophotometer; 177 Holliston, MA, US). Acarbose was used as control, following the same protocol. To calculate the percentage inhibition of α‐amylase activity, the Equation 1 was applied. To determine the sample concentration (extract/acarbose) responsible for inhibiting 50% of the α‐amylase activity, the IC₅₀ value was calculated according to the logarithmic regression equation model based on the plotted data.

where I is the percentage of inhibition activity, A_control is the absorbance value obtained when the sample dilution is replaced with PBS (representing 100% enzyme activity), A_sample is the absorbance value obtained from the tested diluted extract solution. The logarithmic regression equation model based on the plotted data was applied to determine the IC₅₀ (mg GAE/mL) of extract and acarbose (mg/mL).

To analyze the inhibitory activity of the extract on α‐glucosidase, the protocol described by Serea et al. (2022) was used with minor modifications. A volume of 50 μL of α‐glucosidase from Saccharomyces cerevisiae (1 U/mg in 0.1 M PBS, pH = 6.9) was pre‐incubated with 100 μL of different diluted extract solution (concentration expressed in mg GAE/mL) at 37°C for 10 min. By the addition of 100 μL substrate, 25 mM p‐nithophenyl‐α‐D‐glucopyranizide and 1600 μL 0.1 M PBS, pH = 6.9, the enzyme reaction was initiated. After incubation at 37°C for 30 min, 800 μL of 0.2 M Na₂CO₃ was added to stop the enzymatic reaction. The absorbance of the samples was measured at 405 nm using a UV–VIS spectrophotometer (Biochrom; Libra 22 UV/Visible Spectrophotometer; 177 Holliston, MA, US). Acarbose was used as control. Equation 1 was used to calculate the percentage inhibition, and the logarithmic regression equation model based on the plotted data was applied to determine the IC₅₀ (mg GAE/mL) of extract and acarbose (mg/mL).

To determine the inhibitory effect of the extracts on lipoxygenase, the protocol described by Serea et al. (2022) was used, with some modifications. A volume of 50 μL of enzyme solution (1 mg/mL lipoxygenase from Glycine max (soybean), dissolved in PBS 0.1 M, pH = 9.0) and 50 μL of different diluted extract solution (concentration expressed in mg GAE/mL) were incubated for 10 min at 37°C. Subsequently, 50 μL of substrate (0.05 mM linoleic acid, dissolved in PBS 0.1 M, pH = 9.0) was added, and the mixture was incubated for 20 min at 37°C. Finally, 2 mL of PBS 0.1 M, pH = 9.0 were added to the reaction mixture. Absorbance was measured at 234 nm using quartz cuvettes and a UV–VIS spectrophotometer (Biochrom; Libra 22 UV/Visible Spectrophotometer; 177 Holliston, MA, US). The inhibition was calculated based on Equation 1. Quercetin was used as control, and the results were expressed in IC₅₀ values (mg GAE/mL).

The intrinsic fluorescence emission spectra were obtained using a LS‐55 Luminescence Spectrometer (Perkin Elmer, Waltham, MA, USA), as described by Aprodu et al. (2020). In a typical experiment, 0.1 mL of 5 mg/mL of enzyme solutions in phosphate buffer (0.1 M) at different pH values, according to the optimum activity of the enzymes (7.4 for α‐amylase, 6.9 for α‐glucosidase and 8.0 for lipoxygenase), was placed into the cuvette. A volume of 2 mL (α‐amylase and lipoxygenase) or 3 mL (α‐glucosidase) of phosphate buffer at corresponding pH was added, followed by manual titration by successive additions of extract stock solution (increasing concentration from 0 to 1.2310⁻⁶ Mol/L GAE). The fluorescence intensities were corrected for dilution effects before they were considered for data analysis.

The fluorescence quenching mechanism was described by the Stern‐Volmer equation, and the quenching constants of experimental data were obtained by regression analysis (Aprodu et al. 2020). The binding constants (k_b), the number of binding sites (n), and thermodynamic parameters were estimated according to Dumitraşcu et al. (2016).

Plastic tubes (1 cm diameter) were filled with 0.15 mL of enzyme solutions. The samples were heated at temperatures of 37°C, 45°C, 50°C, and 55°C for 10 min using a thermostatic water bath (Digibath‐2 BAD 4, Raypa Trade, Barcelona, Spain). The heated solutions were then cooled by introducing the tubes in ice water, such as to avoid further thermal denaturation.

The molecular docking tests were further used to check the binding particularities of the main polyphenols identified in the red grape pomace, by the pro‐inflammatory and metabolic syndrome‐associated enzymes. The molecular docking tests were carried out by means of AutoDock Vina on SwissDock (Eberhardt et al. 2021; Bugnon et al. 2024). The refined α‐amylase (6Z8L.pdb; Axer et al. 2021), α‐glucosidase (5NN5.pdb; Roig‐Zamboni et al. 2017), and lipoxygenase (3O8Y.pdb; Gilbert et al. 2011) models, selected from the RCSB Protein Data Bank, served as receptors for the following ligands: vanillic acid, azelaic acid, phlorizin, cis‐polydatin, chlorogenic acid, suberic acid, abietic acid, t‐resveratrol. The ligands used in docking tests were selected based on the chromatography results. For each type of enzyme‐ligand complex, the top scoring model, characterized by the highest affinity, as indicated by AutoDock Vina ranking, was further in‐depth characterized using the PDBePISA tools (Krissinel 2010) and VMD 1.9.3. software (Humphrey et al. 1996).

Microsoft Excel for Microsoft 365, MSO version 2412, build 16.0.18324.20240 (Microsoft, Redmond, WA, USA) was used for statistical processing of the data obtained, and the results were expressed as mean ± standard deviation. The data were statistically processed using the One‐Way ANOVA method, using the Tukey test. Values that do not share the same letter indicate statistically significant differences, with p‐value < 0.05.

The extract obtained by UAE using NaDES as extraction solvent contained 3.80 ± 0.01 mg GAE/mL. A detailed polyphenolic profile was obtained by using U‐HPLC (Table 1). As it can be observed in Table 1, the chromatographic analysis allowed the identification of thirty‐two polyphenolic compounds, predominantly phenolic acids (73%), with a total polyphenol content of 575.58 mg/kg. Regarding the phenolic acids, the main compound was vanillic acid (236.36 mg/kg), followed by azelaic acid (50.09 mg/kg), chlorogenic acid (31.42 mg/kg), suberic acid (30.94 mg/kg), and abietic acid (25.01 mg/kg). As regarding the flavonoids, phlorizin was found in a higher concentration of 47.19 mg/kg, whereas the main stilbenes found were polydatin (32.05 mg/kg) and t‐resveratrol (20.66 mg/kg). From the data reported, it seems that the NaDES variant used in this study allowed us to enhance the extraction of phenolic acids when compared with flavonoids and anthocyanins. Therefore, when using HPLC technique, Balan et al. (2025) identified only nine bioactives, namely phenolic acids (gallic acid, caffeic acid, protocatechuic acid, ellargic acid, and chlorogenic acid), flavonoids (epigallocatechin, catechin, and rutin trihydrate), and terpenoids (cafestol). Epigallocatechin and catechin were the main compounds with the highest concentration in the NaDES extract, with values of 91.77 ± 0.02 and 4.51 ± 0.01 mg/mL, respectively. Moutinho et al. (2023) reported a concentration of polyphenolic compounds of 2.407 ± 0.032 mg/mL, with phenolic acids in a concentration of 1.212 ± 0.015 mg/mL. These authors identified by HPLC–DAD–ESI–MS/MS in extracts obtained at 60°C using 3.5% HCl and 70% ethanol from the variety “Touriga Nacional” and “Sousão” fourteen compounds among which were found: 3‐O‐caffeoylquinic acid, p‐coumaroylquinic acid, 5‐O‐caffeoylquinic acid, apigenin‐O‐pentoside, quercetin‐3‐O‐rutinoside, myricetin‐O‐rutinoside, quercetin‐3‐O‐glucoside, kaempferol 3′,4′‐di‐O‐rhamnoside, quercetin‐O‐pentoside, delphinidin‐3‐O‐glucoside, peonidin‐3‐O‐glucoside, malvidin‐3‐O‐glucoside, petunidin‐3‐(6′coumaroyl)‐glucoside. The presence of four anthocyanins in the extract was suggested, whereas in our study, no anthocyanin was detected.

SwissADME and ProTox‐3.0 bioinformatic tools were employed to assess the absorption, distribution, metabolism, and excretion (ADME) as well as the oral and nutritional toxicity of the phytochemical compounds identified in the NaDES extract. This analysis aimed to predict and estimate the potential of the extract as an ingredient in functional food products. The theoretical and computational predicted models and parameters are presented in Figure 1, Table 2.

The physicochemical properties of the investigated compounds (Table 2), obtained from SwissADME, indicate a diverse structural profile within the extract. The molecular weights of these compounds range from 138.12 g/mol for 4‐dihydroxybenzoic acid to 610.56 g/mol for hesperidin, highlighting the presence of both low and high molecular weight phenolic compounds. The majority of the investigated compounds (90.70%) have a molecular weight below 500 g/mol, which, according to Lipinski's Rule of Five, suggests they are likely to exhibit good oral bioavailability in humans (Varma et al. 2006). Additionally, the topological polar surface area (TPSA) values varied significantly, ranging from 37.30 Ų for abietic acid to 269.43 Ų for rutin, reflecting considerable differences in molecular polarity. Furthermore, 72.09% of the compounds showed a TPSA of lower than 140 Ų and < 10 rotatable bonds. According to Veber's Rules, these characteristics suggest that these compounds are likely to demonstrate good oral bioavailability, often providing a better predictive filter than Lipinski's Rule of Five (Soares et al. 2023).

The BOILED‐Egg (Brain Or IntestinaL EstimateD) permeation predictive model, shown in Figure 1, visually illustrates the human intestinal absorption (HIA), represented by the white region of the model, and the blood–brain barrier (BBB) penetration, represented by the yellow region (Daina and Zoete 2016). Only 23.23% of the investigated compounds were predicted to penetrate into the BBB. These compounds include 4‐dihydroxybenzoic acid, p‐coumaric acid, ferulic acid, abietic acid, azelaic acid, sebacic acid, pinocembrine, chrysin, trans‐resveratrol, and isorhapontigenin, suggesting that these compounds may exert neuroprotective or cognitive‐enhancing effects within the central nervous system. Conversely, limited BBB penetration provides an additional safety margin from a food application perspective by minimizing unintended central nervous system exposure (Obermeier et al. 2013). Additionally, 25.58% of the compounds, including 3,4‐dihydroxybenzoic acid, chlorogenic acid, rosmarinic acid, vitexin, naringin, hesperidin, myricetin, rutin, phlorizin, procyanidin B2, and malvidin‐3‐O‐glucoside, are positioned outside the HIA region and are predicted to demonstrate low gastrointestinal (GI) absorption with bioavailability scores between 0.11 and 0.56. These findings align with published literature reporting that flavonoid compounds generally require enzymatic deglycosylation or microbial metabolism before absorption in the small intestine or colon (Németh et al. 2003; Braune and Blaut 2016). Most of the compounds (81.40%) are predicted to be non‐P‐glycoprotein (P‐gp) substrates, with the exception of catechin, epicatechin, hesperidin, myricetin, rutin, phlorizin, and polidatin (cis and trans isomers), which are predicted to be potential P‐gp substrates. The P‐gp substrates compounds are subject to cellular efflux, making them less likely to exert local intestinal biological activity (Varma et al. 2006).

The ProTox‐3.0 toxicity predictions (Table 2) indicate that the investigated compounds have a safe profile. The lowest predicted median lethal dose (LD50), which is the amount at which 50% of test subjects die from exposure to a compound, was estimated at 159 mg/kg body weight for myricetin and quercetin, with 100% accuracy. This level falls within the III toxicity class, indicating that at concentrations between 50 < LD50 ≤ 300 mg/kg body weight, these compounds are considered toxic if swallowed. However, these predictions should be interpreted in the context of exposure levels, as the concentrations typically found in dietary intake are substantially lower than those associated with toxicological concern. Most of the other compounds are classified within toxicity classes IV to VI, indicating low acute toxicity to non‐toxicity and supporting a favorable safety profile. Furthermore, none of the investigated compounds fell within the highly toxic categories (I and II), suggesting no major acute toxicity concerns for food‐based applications under typical dietary exposure conditions.

Nutritional toxicity predictions showed predominantly negative outcomes for the phytochemical compounds, whereas several compounds (37.21%) were predicted to be potential nutritional toxic, with moderate probability ranging from 0.51 to 0.71. These in silico predictions highlight the importance of dose‐dependent evaluation and confirmation through experimental safety assessments.

From a food application perspective, the NaDES extract shows several advantageous characteristics related to the investigated compounds. The physicochemical profile of the identified phytochemical compounds indicates structural diversity that enables complementary biological effects, which are favorable for GI absorption. The polar compounds may exhibit reduced direct permeability but potential gut‐localized functionality. The coexistence of highly and poorly absorbed compounds could lead to dual functionality, such as rapid systemic antioxidant and anti‐inflammatory activity alongside sustained GI and microbiota‐modulating effects. Limited BBB permeability reduces potential systemic safety concerns, and the compounds predominantly fall within low acute toxicity classification predictions.

The extract showed potent inhibition on α‐amylase, allowing us to estimate an IC₅₀ value of 1.12 ± 0.12 mg GAE/mL (Table 3). The corresponding value for acarbose was 0.14 ± 0.01 mg/mL. Therefore, a higher extract concentration is needed for inhibition of 50% of enzyme activity. However, when expressed as inhibitory ratio (%) (Equation 1), a 99% inhibition was reached when adding 1.52 mg GAE/mL, suggesting a high inhibition rate.

In order to gain more structural details about small biomolecules binding to the α‐amylase, important information on the binding parameters and the change in the microenvironment of the fluorophore in the macromolecules can be provided by means of intrinsic fluorescence quenching. The approach is based on the sensitivity of tryptophan and tyrosine fluorescence properties to the polarity of the specific local environment (Vivian and Callis 2001). Therefore, upon successive titration of the enzymatic solution with extract in concentrations ranging from 0 to 1.2310⁻⁶ Mol/L GAE, a significant 18 nm red‐shift was found, suggesting the unfolding of the quaternary and tertiary structure of the enzyme, induced by the binding of small ligands. In general, red shifts in λ_max are associated with the extensive exposure of the previously buried hydrophobic residues, such as Trp and Tyr to the solvent molecules. Additionally, increasing the concentration of the extract up to 1.2310⁻⁶ Mol/L GAE caused the reducing of α‐amylase fluorescence, which was interpreted as the small bioactive from extract bound to the protein (Figure 2a).

In general, when heating, the λ_max red‐shifted with 11 nm, suggesting that the enzyme structure unfolds, thus becoming more prone for binding. The heat‐ and binding particularities of the enzyme molecule behavior allow us to divide the fluorescence quenching mechanism into static quenching and dynamic quenching, based on the binding constant values (Table 3). From Table 2 it can be observed that the K_SV values decreased with increasing temperature, from (1.57 ± 0.50) × 10⁶ L/Mol at 22°C to (1.02 ± 0.08) × 10⁶ L/Mol at 55°C. Regardless of the temperature, the Stern‐Volmer constants were significantly lower than the maximum value for the diffusion‐limited collisional quenching (k_q) (2.0 × 10¹⁰ M⁻¹ s⁻¹) (Lakowicz 2006), thus indicating that the interactions between α‐amylase and RGP bioactive is a static, enthalpy‐driven process (Callis 2014) (Table 4).

In case of a static mechanism, the fluorescence data were analyzed to estimate the binding constant (K_b) and the stoichiometry number (n) values for the complexes (Table 3). From Table 3, it can be observed that, K_b values decreased with increasing the temperature, whereas n values were around one, indicating the presence of at least one binding site able to bind near the hydrophobic sites of the enzyme. The corresponding K_b values were about 1.49 ± 0.42 × 10⁶ L/Mol at 22°C and 1.05 ± 0.11 × 10⁶ L/Mol at 55°C. Different values for binding constants are reported, considering the ligand‐protein structural relationship. For example, Aprodu et al. (2020) estimated significant higher values for the binding interaction between folic acid and bovine lactoferrin, suggesting values of 0.28 ± 0.01 × 10⁸ L/Mol for K_SV and 1.68 ± 0.01 × 10⁸ L/Mol for K_b, respectively.

The intermolecular forces involved in the ligand‐protein complex formation may include hydrophobic forces, van der Waals interactions, electrostatic interactions, and hydrogen bonds (Ross and Subramanian 1981). To distinguish the contribution of each intermolecular energy to the complex formation, the standard Gibbs free energy change (ΔG), the standard enthalpy change (ΔH), and the standard entropy change (ΔS) of interactions were estimated, as described by Dumitrașcu et al. (2018). From Table 4 it can be seen that the binding mechanism of RGP bioactive to α‐amylase is complex, involving hydrophobic bonding, as indicated by the positive value of ΔH and ΔS. The kinetic and thermodynamic binding values indicated that the conformational selection model, as described by Du et al. (2016) can describe the enzyme‐ligand binding mechanism.

When analyzing more in depth, it seems that the native protein does not exist as a rigid conformation, but rather an ensemble of flexible conformational states (flexible fit), which coexist in equilibrium, so that ligands can selectively bind to the most suitable conformation, with the equilibrium shifting towards that binding state. However, as explained by Ross and Subramanian (1981), it is possible that the binding between α‐amylase and polyphenols from the extracts to follow a two‐step binding mode, involving van der Waals interactions and hydrogen bonding and partial immobilization, followed by hydrophobic interactions, leading to an inhibitory effect. Our results on the NaDES extract interaction with α‐amylase, a protein with a particular pharmacological interest, suggest that the polyphenols from RGP can act as antidiabetic molecules.

Additional details on the mechanism involved in the inhibition exerted by the phenolic compounds extracted from red grapes pomace on the α‐amylase activity were collected by analyzing the enzyme‐ligand complexes obtained from molecular docking tests. The enzyme appeared to accommodate most of the tested ligands, with rather low affinity, in a shallow cavity located on the surface of the molecule, the free interaction energy values ranging from −0.2 to 1.7 kcal/mol (Table 5, Figure 3). The only exception concerned vanillic acid, which exerted high affinity (ΔG_int of −0.8 kcal/mol) towards the substrate binding site. In particular, vanillic acid directly interfaced the Asp¹⁹⁷, Glu²³³, and Asp³⁰⁰ residues from the active site, which are pivotal for the catalytic activity of human pancreatic α‐amylase (Axer et al. 2021). The complex formed by α‐amylase with vanillic acid appeared thermodynamically stable (ΔG_diss of 2.8 kcal/mol), suggesting that no spontaneous detachment of the molecules might occur. The hydrogen bonds with Arg¹⁹⁵ and Glu²³³, ionic interaction with His³⁰⁵, hydrophobic contacts with Trp⁵⁸, and π‐staking interaction with Tyr⁶² are mainly involved in the stabilization of the interaction surface. Our results regarding the potential interference of vanillic acid with the catalytic activity of α‐amylase comply with the observations of previous studies investigating the inhibitory mechanism of various phytochemicals. In particular, malvidin‐3‐glucoside (Serea et al. 2022) and gallocatechin gallate (Wu et al. 2018) were previously reported to overlap the active site of the enzyme, with downstream effect on hydrolytic activity.

When comparing with α‐amylase, the NaDES extract showed a significantly higher inhibitory potential on α‐glucosidase, with an IC₅₀ value of 0.47 ± 0.03 mg GAE/mL (Table 3), whereas the corresponding value for acarbose was 0.72 ± 0.01 mg/mL. Therefore, a lower extract concentration is needed for the inhibition of 50% of enzyme activity when compared with acarbose. The extract was able to inhibit 99% of α‐glucosidase activity at a concentration of 1.90 mg GAE/mL. The results suggest that the extract shows potent antidiabetic activity.

The fluorescence emission spectra of α‐glucosidase at 22°C treated with increased concentration of extract are shown in Figure 2b. Similarly, with α‐amylase, given the gradual addition of RGP extracts, the fluorescence intensity of α‐glucosidase gradually decreased. Therefore, increasing the TPC concentration from 0 to 1.23 × 10⁻⁶ Mol/L GAE, the relative fluorescence intensity gradually decreased from 100% to ~61% (Figure 2b). Heating caused unfolding of the enzyme conformation, with a maximum red‐shift of 5 nm, from 340 nm at 22°C to 346 nm at 55°C. When quenching, the maximum extent of 6.5 nm red‐shift in λ_max was found, suggesting the unfolding of polypeptide chains, allowing the exposure of previously buried hydrophobic amino acids.

The Stern‐Volmer and binding constants are given in Table 3, together with the number of binding sites. At 22°C, the binding constant value were of 1.18 ± 0.04 × 10⁶ Mol/L, whereas increasing the temperature lead to an initial decrease, to 1.05 ± 0.01 × 10⁶ Mol/L at 37°C, followed by an increase up to 1.50 ± 0.14 × 10⁶ Mol/L at 45°C and further decreases. The increase–decrease of the K_b values may suggest a temperature mediated flexible fit between the enzyme and polyphenols from the extracts. Anghel et al. (2023) observed the same trend in the binding constant values as depending on temperature, suggesting significantly lower value of 25.09 ± 2.14 × 10⁻² Mol/L for the GP extract obtained by S/L UAE using 70% ethanol. These differences may be due to the different extraction conditions, allowing the increase in the anthocyanin's content in the extract. The value of binding sites suggested that polyphenols from the extracts had one inhibition site or a single class of inhibition site on α‐glucosidase. Based on the positive values of ΔH and ΔS, it may be assume that the major interactions involved in protein‐ligand binding are hydrophobic. When comparing the kinetic and thermodynamic parameters, it seems that the small biomolecules from RGP showed a higher affinity for α‐glucosidase when compared with α‐amylase, whereas the binding process was enthalpy driven (Table 4).

The ability of the major phenolic compounds extracted from the red grapes to recognize and bind to different sites on the α‐glucosidase molecule was also tested through the in silico investigations. Analyzing the results presented in Table 5, Figure 3 one can observe that the ligands preferentially attached to three different cavities located on the protein surface. Out of the tested ligands, of particular interest is suberic acid, which interacts with the catalytic GH31 domain of the α‐glucosidase molecule, being in direct contact with Asp⁵¹⁸ and Asp⁶¹⁶ residues, which are responsible for the catalytic nucleophile and acid/base activity of the enzyme (Roig‐Zamboni et al. 2017; Serea et al. 2022). This in silico observation supports the experimental results regarding the inhibitory potential of the phenolic compounds from red grape pomace on α‐glucosidase activity, suggesting the antidiabetic activity of suberic acid. In addition to the good affinity between molecules, as indicated by the ΔG_int of −1.1 kcal/mol, the α‐glucosidase—suberic acid complex appeared stable from the thermodynamic point of view (ΔG_diss of 2.2 kcal/mol).

Roig‐Zamboni et al. (2017) provided atomic level details on the high‐resolution models of α‐glucosidase in complex with different active site directed iminosugar inhibitors, such as 1‐deoxynojirimycin and its N‐hydroxyethyl‐ derivative, which are pharmacological chaperones. They explained that the complexes were stabilized through hydrogen bonds involving the side‐chains of Asp⁴⁰⁴, Asp⁵¹⁸, Arg⁶⁰⁰, Asp⁶¹⁶, and His⁶⁷⁴ and hydrophobic contacts with Trp³⁷⁶, Ile⁴⁴¹, Trp⁵¹⁶, Met⁵¹⁹, Trp⁶¹³, and Phe⁶⁴⁹ (Roig‐Zamboni et al. 2017). Our results are in good agreement with the experimental observations of Roig‐Zamboni et al. (2017), as all amino acids involved in the interaction with 1‐deoxynojirimycin were found in the present in silico experiment to be in direct contact with suberic acid (Table 5).

Our results are as well supported by Panigrahy et al. (2020), who reported that suberic acid is a major compound found in the Hedychium coronarium Koen. extract, which exerts high inhibitory activity towards α‐amylase and α‐glucosidase. Moreover, Ofosu et al. (2022) showed that fermentation with Pediococcus acidilactici of the pressurized cooked sorghum increased the level of suberic acid, while significantly enhancing the α‐glucosidase inhibition.

As presented in Figure 3b, except for suberic acid, all other ligands are either accommodated by a pocket of the N‐terminal β‐sheet domain (vanillic acid and azelaic acid) or by a cavity at the confluence of N‐terminal β‐sheet and GH31 domains (phlorizin, cis‐polydatin, chlorogenic acid, abietic acid, and resveratrol). Although the attachment of these ligands to the enzyme surface does not impede substrate binding, as no direct participation of the catalytic amino acids was observed, some local structural rearrangement might occur; therefore, potentially interfering with the hydrolytic activity.

The effect of RGP extract on lipoxygenase fluorescence intensity is shown in Figure 2c, which indicates the shift in enzyme emission maximum when adding different concentrations of extract. Regardless of the temperature, lipoxygenase showed a λ_max of 339 nm, with no significant differences in fluorescence intensity. On titration with RGP extract, the fluorescence intensity decreased regularly with increasing the concentration of the extract, due to a variety of molecular interactions, such as molecular rearrangements, excited‐state reactions, ground state complex formation, energy transfer, and collisional quenching, indicating interactions between bioactive compounds from the extract and lipoxygenase.

When quenching, significant red‐shifts were observed in a temperature dependent manner. The increase in temperature from 22°C to 37°C led to less flexible molecules, with red‐shifts of 6 and 4 nm, respectively, suggesting local rearrangements in protein structure which led to a lower affinity for ligands. Increasing the temperature up to 55°C followed by quenching led to significant red‐shifts of 12 nm at 45°C, 7.5 nm at 50°C, and 10.5 nm at 55°C, indicating a more heat‐induced flexible molecule, prone to bind small molecules from the extract.

The intrinsic fluorescence data were correlated with the inhibitory capacity of the extract. Therefore, the IC₅₀ value was 0.06 ± 0.01 mg GAE/mL, whereas for quercetin the corresponding value was significantly higher at 5.17 ± 0.07 μg/mL. However, the extract showed a high inhibition rate, a concentration of 0.4 mg GAE/mL leading to the inhibition of about 82% of the initial enzyme activity.

The binding parameters for RGP to lipoxygenase are given in Table 3. Two mechanisms of binding between small molecules and proteins are possible, according to the K_SV constant dependence on temperature: a static and dynamic quenching mechanism, respectively. For the static quenching, quenching rate constants decreased with increasing temperature, but the contrary effect was seen in the case of dynamic quenching (Du et al. 2016). Based on data shown in Table 3, the mechanism of binding is static up to 37°C and dynamic at higher temperatures. From the value of n (Table 3), it was found that there is at least one independent class of binding sites on lipoxygenase for the extract bioactive, which indicates that more than one extract molecule is bound to one enzyme molecule.

The K_b values followed a temperature induced decrease–increase trend, which means that the interactions between polyphenols from the extract and lipoxygenase are temperature mediated, suggesting a flexible fitting between molecules. The thermodynamic parameters indicated the involvement of van der Waals forces and/or hydrogen bonds in the temperature range of 37°C–50°C, due to the negative value of ΔH and ΔS (Table 4).

The molecular docking results involving the lipoxygenase as a receptor for the major phenolic compounds from the red grape pomace indicated that the tested ligands attach with different affinities to three different sites of the enzyme (Figure 3c, Table 5). The complexes formed by lipoxygenase with suberic acid and resveratrol are of particular importance, as the ligands are able to recognize and interact with amino acids of the active site and the catalytic iron. The particular attachment of any of the two ligands obstructs the access channel to the catalytic iron, as suberic acid is in direct contact with Phe¹⁷⁷, while resveratrol with Tyr¹⁸¹, whose side chains are positioned inward the cavity (Gilbert et al. 2011). Both ligands are in close vicinity and each interact with three of the four amino acids engaged in iron coordination, namely His³⁶⁷, His³⁷², His⁵⁵⁰, and Ile⁶⁷³ (Table 5). Moreover, suberic acid and resveratrol established hydrophobic contacts with the Leu⁴¹⁴ residues, which controls the O₂ access in the vicinity of the substrate (Gilbert et al. 2011). These observations imply the direct involvement of the suberic acid and resveratrol in hampering the substrate transformation by lipoxygenase.

Although the other ligands do not bind in the vicinity of the active site, their contribution to enzyme inhibition should not be neglected, even at low concentration. Because of the intrinsic instability of lipoxygenase (Gilbert et al. 2011), even small spatial reorganization of the cavities accommodating the ligands might result in alteration of the catalytic activity. Indeed, Mahesha et al. (2007) reported the non‐competitive inhibitory activity of soy isoflavones, which attach to different sites of lipoxygenase compared to the substrate. Mahesha et al. (2007) also suggested a second lipoxygenase inhibition mechanism, consisting of the ability of the phenolic acids to scavenge the free radicals, a process which provides electrons to the Fe³⁺, turning the enzyme to the resting state.

Data are contained within the article.